Application of the Cascade Theory to Calculation ... - ACS Publications

Application of the Cascade Theory to Calculation of Particle Scattering Factors of Polydisperse Systems of Stiff Chains. I. Franken, and W. Burchard...
0 downloads 0 Views 819KB Size
848

Franken, Burchard

Macromolecules

Application of the Cascade Theory to Calculation of Particle Scattering Factors of Polydisperse Systems of Stiff Chains I. Franken and W. Burchard* Institute of Macromolecular Chemistry, Lhiuersity Received March 26. 1973

of

Freiburg, 78 Freiburg i.Br., West Germany.

ABSTRACT: The mean-square radii of gyration (Sz), and the particle scattering factors P,(e) of polydisperse (M,/M, = 2) stiff linear chains, the weight-average molecular weights, and the weight distributions were calculated applying the cascade theory of Good and of Gordon. Making use of the well-known result of Schulz that polydispersity is decreased by chain coupling, the mentioned averages were derived for lower degrees of polydispersity (M,/M, < 2). Relatively simple analytic expressions were obtained even when chain stiffness was taken into account. Increasing chain stiffness and decreasing polydispersity both cause similar deviations of the meansquare radius of gyration from the Gaussian behavior at a low degree of polymerization DP,. Thus, according to the properties of the mean-square radii of gyration, polydisperse chains seem to have a lower chain stiffness than the chains have in reality. Conversely, increasing chain stiffness and increasing polydispersity cause a similar downturn of the reciprocal particle scattering factor a t large angles. Therefore polydisperse chains give indications of a higher chain stiffness, if the angular dependence of the scattered light is analyzed.

1. Introduction

In the course of our theoretical studies on branched polymers the correlation between molecular weight distribution and conformation of linear and branched macromolecules appeared to us to be of particular interest. As is well known, polymerization of linear chains results in a more or less broad molecular weight distribution depending on the kinetic mechanism.1-4 Furthermore, decreasing molecular heterogeneity is achieved by multiple coupling of chain^.^ While the mechanism of chain coupling has no influence on the width of the molecular distribution, it certainly affects the conformation of the macromolecules. To give an example, the molecules of head-to-tail coupled linear chains will exhibit more extended coil dimensions than the corresponding star-like polymers, whose primary chains are coupled a t one point. The purpose of this paper is the calculation of conformational z-averages of the mean-square radius of gyration (SZ), and the particle scattering factor P,(@),as well as .the already-known molecular weight averages M, and M,. While the molecular weight averages can be calculated by applying the simple convolution principle, a more involved statistical method is required for the calculation of the conformational averages. The cascade theory, developed by G o ~ d and ~ - ~by Gordonlo-12 for branched polymers, provides a suitable tool for such calculations, also for linear molecules. The essential features of cascade theory is the use of probability generating functions, which by a differentiation operation yield the desired conformational averages. I t is of great advantage that these averages are simultaneously performed with respect to a molecular weight distribution. By combination of existing theories of homodisperse stiff c h a i n ~ l ~ -with ~ S the cascade theory the influence of v6

G. V. Schulz. Z. Phys Chem. (Leiprig),30.379 (1935). P . J. Flory, J. Amer. Chem Soc., 38. 1877 (1936). P . J. Flory, ;I. Amer. Chem Soc , 64.2205 (1942). P . J. Flory,J. Chem. Phys., 12,425 (1944). (5) G. V. Schulz, Z. Phys. C h e n . (Leiprig).43.25 (1939). (6) B. H. Zimm, J . Chem. Phys., 16,1099 (1948). (7) I. J . Good, Proc. Cambridge Phil. Soc. 43.360 (1948). ( 8 ) I. J . Good, R o c . Cambridge Phil. Soc , 56,36 (1960). (9) I. J . Good, Proc. Ro?., Soc., Ser. A , 272. 54 (1963). (10) M. Gordon, Proc. Roy. S o c . . Ser. A , 268, 240 (1962). (11) G. R. Dobson and M. Gordon, J. Chem. Phys., 41.2389 (1964). (12) D. S. Butler, M. Gordon, and G. N . Malcolm, Proc. Roy. Soc., Ser A , 295.29 (1966). (13) A . Peterlin, Makromoi. Chem., 9,244 (1952). (14) S. Heine, 0. Kratky, and R. Roeppert. Makromol. Chem., 56, 150 (1962). (1) (2) (3) (4)

molecular heterogeneity on the scattering behavior of stiff chains is accessible to calculation. For our purpose, Daniels’ approach’s to a solution of the chain stiffness problem is most appropriate, because the resulting Fourier transform of the end-to-end distribution (needed for the calculation of particle scattering factors) can be directly introduced into the cascade theory. In the following, analytical expressions are derived for the head-to-tail coupling. Details of the calculation for star-like coupling will be given elsewhere.20

2. Linear Chains with the Most Probable Chain-Length Distribution For a better understanding of the method used below the simple case of linear chains with a most probable distribution of the molecular weight (M,/Mn = 2) is treated in detail. 2.1. Principles of Averaging Due to the Cascade Theory. For the calculation of conformational averages the following double sum has to be considered.21

$t,

is a weighting function depending on the distances

rLl between two chain elements (repeat units) i and j ; the

angular brackets indicate the corresponding average. If a suffix is added, a further average over a molecular weight distribution is performed. A simple counting system for the calculation of such double sums is obtained by considering the polymer molecule as a “family” tree or more precisely as a rooted tree where an arbitrary chain element is choosen as a root. Obviously each chain element has the same chance of becoming a root of a tree and thus a full forest of x isomorphic trees has to be considered for a chain consisting of x monomers or chain elements. Similarly, as is done with family trees, different generations can be introduced for these trees, counting the root as the zeroth generation and P . J. Flory and R. L. Jernigan, J. Amer. Chem. Soc . 90. 3128 (1968). Y. Tagami. Macromolecules, 2 , 8 (1969). P . Sharp and V. A. Bloomfield, Biopolymers. 6.1201 (1968). W. Burchard and K . Kajiwara, Proc. Ro?. Soc., Ser. A , 316. 185 (1970). (19) H. E. Daniels, Proc Roy. Soc.. Ser. B, 63. 290 (1952). (20) U’.Burchard. Manuscript in preparation. (21) B. H . Zimmand W.H. Stockrnayer,J. Chem Phys., 17.1301 (1949). (15) (16) (17) (18)

Particle Scattering Factors

Vol. 6, No. 6, November-December 1973 denoting the next neighbors to be in the first, second, etc., generations (see Figure 1). In general, owing to the conditions of synthesis, polymers are obtained with more or less broad molecular weight distributions. Let N(x) be the frequency distribution of a linear chain of the degree of polymerization x, then the average degree of polymerization is found by taking the average with respect to x N ( x ) , because the whole assembly of trees has to be considered. Therefore the average obtained is the weight-average degree of polymerization DP,. All further statistical problems are simplified by introducing generating functionsZ2 instead of using probability distribution functions themselves. Following Gordon10-12 such generating functions can be calculated from the probabilities of reaction, sometimes called the link probabilities. The link probability generation function in the zeroth generation for linear chains is given by

F,(e)

=

(1

-

cy

+

ae)2

F,(O)

=

1 -

(Y

+

a0

peating unit), Mw = the weight average of the molecular weight of the chain, (Sz), = the z average of the meansquare radius of gyration, ( r n 2 ) = the mean-square distance between a monomer of the chain in the nth generation and the root; h = (4x/X) sin 812.

Equation 11 is the Fourier transform of the spatial distance distribution W(r), 0 the scattering angle, and X the wavelength of the light in the medium. Degree of Polymerization and the Molecular Weight. The weight average of the degree of polymerization and the molecular weight follow from eq 7 and 8 by summing the corresponding infinite series of eq 6

DP, = (1 M , = MJ(1

(2)

For all subsequent generations (neglecting substitution effects), one has

(3)

In eq 2 and 3 the parameter 1 - a is the probability that a reactive group has not reacted in the course of a reaction and a is the probability, that it has. Since for linear chains the root has two functional groups, the link probability generating function is found by the product of the generating functions of these two groups. In all the other generations only one group is left for further reactions. For determination of the “mean structure of the whole tree” the different link generating functions have to be joined. This can be done by a cascade substitution according to G ~ o d ~ - ~

849

+

@)/(I

+

a)/(l - a)]

-

(12)

cy)

(13)

Mean-Square Radius of Gyration. For the calculation of the mean-square radius of gyration (S2), and of the particle scattering factor P J 8 ) a relationship is required, on the one hand, between the distances ( r n ) , and, on the other hand, the bond length b, the valence angle y, and the number of repeat units n. According to Eyring25 the following relationship holds for the end-to-end distance of the freely rotating chain with a fixed valence angle (rn2) = b’

+

1 cos y 2 cos y 1 - cos y n - (1 - cos Y)’(1 - cosi‘ y)]

(14) Equation 14 is valid for flexible chains (cos y = 0) as well as for stiff ones (cos y > 0).The stiffness of a chain can be characterized by either the persistence length as introduced by Porod26J7

Inserting eq 2 and 3 into eq 4, one obtains

u = b/(l

-

(15)

y)

COS

or by a reduced parameter q

b l a = 1 - cos y (15’) For very long chains ( n >> 1, Gaussian region), eq 14 q =

where is a weighting function, now dependent on the distance of the units in the nth generation from the root. Uo(0) is called the path weighting generating function. Differentiation with respect of 0 a t 0 = 1 yields the averages of the weighting function.

simplifies to

( r n 2 )= b2[(1 t cos y ) / ( l - cos ~ ) ] n (16’)

= b 2 [ ( 2 - q ) / q ] n = b,ff2n

Using the effective bond length beff (eq 16’) and the chain stiffness parameter q, eq 14 takes on the equivalent form

n =1

2.2 Special Weighting Functions. According to Kajiwara et al.24averages with physical meaning are obtained by inserting special weighting functions Insertion of this relationship in eq 12 and 6 yields the z average of the mean-square radius of gyration

(1

where Mo = the molecular weight of the monomer (re(22) For definition and properties of generating functions, see W.Feller.23 (23) W. Feller, “An Introduction t o Probability Theory,” Vol. I, John Wiley, New York, N. Y., 1968, Chapter 11. (24) K . Kajiwara, W. Burchard, and M. Gordon, (1970).

Brit. Polyrn. J , 2. 110

-

+

- CY 1 l + a l -

1

___

”)]

41d l - q )

(17)

For long chains, i . e . , a 1, a Gaussian chain behavior is obtained and eq 17 reduces to (25) H. Eyring, PhyP. R e v . , 39.746 (1932). (26) G. Porod, Monutsch. C h e m . , 80. 746 (1932). (27) 0. Kratky and G. Porod, Reel. Truc. Chim. Pays-Bus, 6 8 . 1106 (1949).

850

Franken, Burchard

1

2

Macromolecules

3

L

5

6

7

Figure 1. The seven possible rooted trees of a linear heptamer.

(S?,

beu2[a/(1-

=

(17')

The same result may be obtained by inserting eq 16 into eq 6. Particle Scattering Factor. Up to now the Fourier transform (eq 11) cannot be analytically expressed in terms of structural parameters b and cos y. However, for long flexible chains the distance distribution W(r,) has the well-known form of a Gaussian distribution12

In this limit the Fourier transform of the spatial distance distribution (see eq 11)is

(T)

= exp(-(rn2)h2/6)

= exp(-Xn2/6) = 2"

(19)

where X 2 = h2b,ffz, h = (4r/X) sin 9 / 2 , and 2 = exp( -X2/6). The particle scattering factor of Gaussian chains, obeying the most probable molecular weight distribution, is obtained now from eq 16 and 19.

PA0) =

1 1

~

a 1 + a2 + a 1 - a2

-

(20)

The Fourier transform of stiff chains was calculated by Danielsl8 ~9 to a second-order approximation

2

1

Figure 2. (a, top) Coupling of m primary chains (schematically). The arrows indicate the points of coupling, the figures the number of primary chain. (b, bottom) Three representatives of possible tree structures of the coupled chain. The labeling is as follows. The first number denotes the primary chain considered, the second number the neighbored primary chain or the free end (indicated by the number 0). The numbers beneath the graphs indicate the primary chains the root was chosen from.

two possibilities of reaction: (a) a monomer can have reacted with the probability CY with another monomer building up thereby a primary chain; ( b ) the monomer could be involved in a coupling reaction with the probability 1 - a where a linkage between this monomer and an end group of an already existing primary chain is formed. Strictly speaking, 1 - CY indicates the probability of finding an end of a primary chain before chain coupling. , ~ + I that a chain After chain coupling (1 - C Y ) ~ ~ indicates end of primary chain i is coupled with probability p l , i + l to primary chain i 1. In the present problem we assume p o , ~= pm,o = 0 and pi,i+l = 1 for all other primary chains. Thus the probability 1 - CY indicates the likelihood of a coupling reaction between two primary chains, with the exception of the first and last primary chain bound in the macromolecule where 1 - CY indicates no reaction (free ends). The procedure of calculation is illustrated in Figure 2a,b. Owing to the condition of randomness in the cascade theory, any repeating unit of this coupled chain can become the root of a tree. Depending on the primary chain considered, one has to distinguish between m different groups of tree structures. All trees, the roots of which are monomers of the same primary chain, are collected thereby into one group. Figure 2b exhibits representatives of trees where either a monomer from primary chain 1, primary chain 2, or primary chain j is choosen as the root. In the first case there are two statistically nonequivalent branches; the branch to the left (in Figure 2b) leads to a free end while the branch to the right leads to an end linked up with the remaining part of the coupled chain ( i e . , m - 1 coupled primary chains). If a monomer of the second primary chain is a root, then the left end is connected with the first primary chain and the right end is linked to the remaining part consisting of m - 2 primary chains. In the general case where a monomer of the j t h primary chain furnishes the root, the left end is coupled with the chain fraction of j - 1 primary chains, and the right end is coupled with the remaining fraction of m - j primary chains. There is a one-to-one correspondence between the number of repeat units in the j t h primary chain and the number of trees in the j t h group, since each repeat unit of a

+

where

For this expression the series in eq 6 can be evaluated analytically resulting in

P;(0)=

(22) The equations for the degree of polymerization and the molecular weight (eq 12 and 13) agree with those derived by Flory for most probable molecular weight distributions. Equations 17' and 20, which describe the conformational averages (S2)2 and P2(t)) for Gaussian chains, are in accord with those for branched polymers,24 if the special case of bifunctional repeat units is regarded. The analytical expressions for conformational averages of heterodisperse stiff chains (eq 17 and 22) are derived here for the first time. 3. Chain Coupling

3.1. Some General Considerations. As was done in section 2, the statistics of coupled chains can be treated with the aid of the cascade theory. The difference from the simple case discussed before consists in the occurrence of

I

Vol. 6, No. 6, November-December 1973 primary chain has to be considered as a root. In the following we confine ourselves to homopolymers, i e . , all primary chains are assumed to have the same degree of polymerization and the same composition.28 The procedure of averaging consists now in averaging over all elements of a group followed by averaging over all groups. 3.2. Generating Function. As already pointed out, the two branches emerging from a root are not statistically equivalent. Therefore, a directional dependence of coupling has to be considered for each of the m groups of trees before the path weighting generating function can be set up. This directional dependence will be recognized from the tree graphs given in Figure 2b. Now the path weighting generating function can be set up for each of the m groups. For the units in the zeroth generation the generating function of the two non-alike branches has to be multiplied; this yields

Particle Scattering Factors

851

Generating functions for the right branch

Differentiation of UOwith respect to fl at 0 = 1 yields a general expression for conformational averages (see Appendix II).

Here i, means the ith element of the j t h primary chain, and $i1 + & + . . . i, is the corresponding weighting function for distances of the path length il i2 + . . . i,. The j-fold sum in the curled brackets takes into account all possible distances between two chain elements of the coupled chain; thus it obeys the condition for the calculation of conformational averages given in eq 1. The frequency of the various distances is given by the coefficients The final sum performs the averaging over all the m groups of the trees. Equation 26 reduces to the already-discussed case of a linear chain with a most probable distribution for the noncoupled chain ( m = 1).

+

The generating function of the whole coupled chain is given by (see also footnote eq F24)

L Y ~ J - ~ .

The expressions in the left parentheses represent the left branches of the tree in Figure 2b, while the right parentheses represent the right branches. The meaning of the indices will now he discussed for the example of the j t h generating function. The first subscript of a generating function always denotes the generation. The first index in the bracket indicates the primary chain to which the monomer units of consideration belong. The second index denotes the coupled primary chain to the left, if it is smaller than the first index, while it denotes the primary chain to the right, if it is larger than the first index. ctlJlt,,J-l, represents the probability of a linkage in the first generation between the root of the tree and a monomer of the chain ,I( ] - 1) in the first generation. (1 LY)LJ~(,,,-~, represents the probability of a linkage in the first generation between the root and a monomer of the neighbored primary chain 0 - 1, - 2). A corresponding nomenclature applies to the linkages of the right branch. For the elements of the first, second, . . . nth generation only one branch sprouts instead of the two for an element in the zeroth generation. The generating functions of the nth generation ( n 2 1)are in analogy to eq 23. Generating functions for the left branch

3.3. Special Averages. As was shown in section 1.2 the different averages are obtained by inserting special weighting functions. Degree of Polymerization30

qn

U,'

= 1

=

DP,

Using eq 26 the degree of polymerization is found to be

DP,

= 1

2 " m - J + a + -1 mJ=, 1 - a

m+cu -~ 1 - a

(27)

Mean-Square Radius of Gyration

qn

UOt= 2DP,(S2),

= (rn2)

Here we have to distinguish between the two cases of very long and flexible chains (the region of Gaussian behavior) and of stiff chains. For Gaussian chains the weighting function is

qPn=

(28) For block copolymers the path weight generating function is given by

(rn2)= berfin

Thus, for coupled Gaussian chains the mean-square radius of gyration is obtained as Here C, is the mass fraction of the j t h primary chain of a special chemical composition, and UO,is the corresponding generating function of the j t h primary chain. If all primary chains have the same average length and the same chemical composition, we have C , = l / m for alll's. (29) p. ch. = primary chain.

(30) According to Schulz5 the number-average degree of polymerization is

DP, = mDP,,,.,t,, where DP,,,,, Ct,., primary chain.

5-

mlil - a)

0'27)

is the number-average degree of polymerization of a

852

Macromolecules

Franken, Burchard *-.-.-.-.-a 19'

Molecular Weight Distribution. Generating functions are also useful for the derivation of probability distribution functions. For instance, the weight distribution function for molecules built up of m coupled primary chains is derived from eq 24 and 25 as follows. Setting 4n = 1, one obtains for the weight generating function the analytic expression

This equation may be written as a power series in terms of

005

ox

00

-

10'

loo

As by definition

102

to3

1o4

W(9) = T w ( x ) 9 "

DP,

0

Figure 3. The z-average mean-square radius of gyration as function of the weight-average degree of polymerization for chains of a certain stiffness (21 monomer units/persistence length) and of various values of heterogeneity. The figures on the curves indicate the number of primary chains per molecule. (The heterogeqeity is M,/M, = 2 for 1 primary chain and M,/M, = 1.01 for 99 primary chains.)

where w ( x ) is the weight distribution function for chains of DP = x, one has

This distribution becomes for x >> m

which approaches the familiar Schulz-Zimm5 ,6 distribution if a N 1 For stiff c h a i n s the corresponding relationship for the weighting function is given by eq 16 when inserted into eq 26.

U,'

=

DP,P,(O)

1

__

DP,

J

x"fl+l

m!

exP(-Yx)

(36)

wherey = 1 - (Y = m/DP,.

+

Again, two different cases have to be distinguished for the Fourier transform: (a) for long G a u s s i a n c h a i n s (see eq 10) the particle scattering factor is found to be

P 2 ( 0 )=

-

4. Computations and Discussion The derived formulas for the linear chains with various heterogeneities are relatively simple and are, a t least formally, valid for all degrees of polymerization. There is a clear correspondence between the number of coupled primary chains, m, and the heterogeneity M,/M,. The latter follows from the equations for the weight- and number-average molecular weights (see eq 27 and F27).

where (S2),Ga"ss is given by eq 28. Particle Scattering Factor sin hr

w(x)

=1

(b) for stiff c h a i n s , on the other hand, the Fourier transforms of the spatial distance distributions are represented by eq 21. Thus, the particle scattering factor of stiff chains is obtained from eq 26.

(37) M,/M, = 1 (cx/m) This result is in agreement with the early findings of Schu1z.l Thus with increasing number of coupled primary chains the polymer becomes more homogeneous. Figure 3 shows the dependence of (Sz),on DP, for a molecule with a persistence length of a = b / q = 21b. Since in all cases (S2), decreases with increasing m values, we found it more instructive to use reduced (S2), values

where (S2),_ is the corresponding value to (S2), a t very large degrees of polymerization (Gaussian region). Figure 4 demonstrates the DP dependence of the fol-

Particle Scattering Factors

Vol. 6, No. 6, November-December 1973

853

10

09 08

no

i

/

07

06 05 04

03 Q2 0.1

I00

IO’

-

lo3

102

IO4

DP,

Figure 4. Dependence of the reduced-mean-square radius of gyration (S2),/(S2),_ on the degree of polymerization for chains of various heterogeneities and flexibilities. Meaning of the figures as in Figure 3; for the definition of the ordinate see eq 33.

53

100

150

200

250

h2(S2JZ

Figure 5. Influence of heterogeneity (chain coupling) on the taverage particle scattering factor of Gaussian chains. The number of primary chains is indicated by the figures on the curves.

scattering data simultaneously provide information on the molecular weight M,, the particle scattering factor P,( 8 ) and the mean-square radius of gyration (S2),. In cascade lowing three cases: (i) a completely flexible chain, l / q = theory the same averages are obtainable as special cases of 1; (ii) a chain with a certain limited flexibility, l / q = 6; a general generating function. Thus, there is a one-to-one (iii) a stiff chain, l / q = 21. While case i has no physical correspondence between physical and theoretical pictures. counterpart, case ii is typical for synthetic polymers like Furthermore, a cascade process is a discrete process and poly(methy1 methacrylate) and polystyrene, and case iii is cascade theory inherently takes account of the discrete found for cellulose derivates. nature of real chain molecules and of the corresponding In all cases characteristic deviations from Gaussian bemolecular weight distribution. In some cases this fact is of havior are observed a t low DP,. These deviations resemimportance for a more detailed description of chain conble the already known behavior of homodisperse stiff chains. Consider a point of deviation where ( S 2 ) z / ( s 2 ) 2 m formations. For instance, the number of residues per turn, together with the persistence length, is an essential pahas reached a value of 0.75; this point migrates from very rameter of helically wound chains. Relationships derived low DP, values to higher ones with increasing chain stiffpreviously for the mean-square distances (rn2) in a disness. A similar, less pronounced shift is obtained with increte homodisperse semiflexible helix31 can directly be increasing m values, l.e., with decreasing heterogeneity. serted in eq 26 when the influence of heterogeneity on the Thus, the curves for polydisperse chains seem to indicate scattering behavior is of interest. A further advantage is a lower chain stiffness than the chain in reality has. given by the fact that the expressions derived by cascade The deviations from Gaussian behavior calculated for ideal, flexible chains, i.e., cos y = 0, appear surprising a t theory are not restricted to long-chain molecules, but are, a t least mathematically, valid for all degrees of polymerfirst sight. These deviations are caused, however, by the ization. effect of the finite chain length, and the same holds also Finally, from the mathematical point of view it seems for stiff chains. The chain stiffness essentially reduces the worth mentioning that this treatment deals with a longnumber of statistically equivalent bonds, and to achieve range correlation (after n generations) by distinguishing Gaussian behavior, a sufficiently large number of statistibetween the two functionalities on both sides of a repeat cally independent bonds are required. unit selected a t random. With respect to a reacting monoThe effect of chain stiffness and heterogeneity on the mer this difference is not evident, but it becomes clear on particle scattering factor is demonstrated in Figures 5 and the consideration of tree structures (given in Figure 2b), 6. Here an increasingly marked downturn occurs with dewhich is the basis of the cascade theory. creasing m values (increasing heterogeneity) and also with increasing chain stiffness. This reflects a well-known feaAcknowledgment. The authors are grateful to Professor ture according to which it may be very difficult to decide M. Gordon for valuable comments. The work was supwhether an observed downturn is due to a high chain stiffported by the Deutsche Forschungsgemeinschaft. ness or due to a considerable heterogeneity. One notices, Appendix I however, that for homodisperse chains the downturn at large angles is preceded by an upturn a t small angles. Weighting Function for the Particle Scattering FacThis is not the case for polydisperse stiff chains. Theretor. The characteristic function 4 ( h , $ ) for the distance fore, accurate measurements a t small as well as at large distribution of two chain elements is expressed by Danangles are necessary for a reliable interpretation of the ielsls as a series of Legendre polynomials scattering data. Some Concluding Remarks. The analytic expressions @(h,IC)= ~ @ s ( h ) P s ( c IC, os (Al) 9 =O derived in this paper appear to us to be of advantage for the following reasons. As is well known, Zimm plots of (31) W Burchard. Brit Polqrn J 3.209, 214 (1971)

854

Macromolecules

Franken, Burchard

+

where is an angle of orientation. The average of $(h,+) taken over all orientations is the weighting function & = (sin hrn/hrn)required for the calculation of particle scattering factors.18 Daniels calculated the two first members of the series in eq A1 and found for the angular independent partlg

60

1

50

For chains with fixed bond angles y the coefficients gl and g2 are Legendre polynomials of the bond angle

g , = cos 1 2

g,

cos2 Y - 1)

= -(3

100

150

2%

200 dtS’),

Figure 6. Influence of heterogeneity on the particle scattering factor of a moderately stiff chain. l / q = 26 is the number of monomer units per persistence length. Meaning of the other figures as in Figure 5.

(A3)

Introducting the persistence length a and a stiffness parameter q

a

= b/(l

- cos Y) - cos Y

q = bja = 1

into eq A2 one arrives a t

11 - q 1 3

+

4(1 1

-

-

h2b2

+

+

q) 3(1 (1 - q),

-

and a similar simplification is possible for the two last sums in eq A7. Thus 4)’ +

J

12 - q

Iterative differentiation of eq 25a yields

h2b2n) (A4)

Ul(1.0;

=

$1

+

This equation can be written in the alternative form

(y)+

C1X2]Z“

=[1

d + C 2 X 4dX2 -(Zn)

(A5)

+

+ a2U3(1,0)’ pan-14n (A10)

aU*,,,o,’ =

$1

a$2

-

n-1

The same result is obtained from the last line in eq 25b m

where the abbreviations were used

Ul(m,m-l)t

c, = -31 410 --q 4 ) 1 (22/3) c, = -

30

- 6q

d 2

-

Can-l$n n-1

u1(1,0;

(All)

Similarly from the second line in eq 25a one finds

+

4) X 2 = h 2 b 2 [ q / ( 2- q)J

q2

(A6)

Z = exp(-X2/6) Appendix I1 Derivation of Eq 26. When differentiating eq 23 with respect of 8 a t 0 = 1,the following relationship is obtained

and from the second last line in eq 25b Ul(m-l,m)’

UI(P,I~’

(A131

Proceeding in eq 25a and eq 25b line by line one finally arrives at

where use was made of eq 24. The two first sums can be simplified

Vol. 6, No. 6, November-December 1973

Conformational Maps for ( S a r ) , and (MeAla),

855

Therefore

Inserting eq A14 and collecting terms with same number of sums this results in eq 26

Comparison of the Conformational Map for Poly(L-proline) with Conformational Maps for Polysarcosine and Poly (N-methyl-L-alanine) Wayne L. Mattice Department of Biochemistry, Louisiana State Uniuersity, Baton Rouge, Louisiana 70803 Received March 23, 1973 ABSTRACT: The presence of the four minima in the conformational map reported by Mark and Goodman for poly(N-methyl-L-alanine)has been confirmed. The conformational map reported by Tanaka and Nakajima for polysarcosine, however, is found to be incorrect due to their failure to consider several crucial interatomic contacts. The conformational maps for both polysarcosine and poly(N-methyl-L-alanine) are found to be sensitive to the orientation selected for the methyl groups. If the geometry of these polypeptides is made similar to that of poly(L-proline), the computations at @ = 120" reveal two regions of low energy, separated by approximately 180" rotation about $. The relative energies of these two minima are extremely sensitive to the geometry selected for the polypeptide, as has previously been found to be the case for poly(L-proline).

A recent treatment2 of the internal dipeptide unit in poly(L-proline) containing peptide bonds in the planartrans conformation is in agreement with the characteristic ratio observed for this polypeptide3 and allows for flexibility of the pyrrolidine ring, which has been demonstrated experimentally by proton4 and 13C5 magnetic resonance studies. Energy minima are obtained near 4, $ = l l O o , 350" and 4, Ji = 120", 130".2,6 The precise location and energy of the minimum near 4, IC, = 120°, 130" is sensitive to the geometry of the pyrrolidine ring.2.7 The energy a t this minimum may be as little as -1 kcal/mol greater than the energy a t 4, $ = 110", 350" if the pyrrolidine ring geometry is suitably chosem2 Conformational maps reported for the internal dipeptide unit in polysarcosine8 and poly(N-methyl-L-alanine),g computed with X N = OO,6 report a low energy for 6 , $ = l l O o , 350°, but do not exhibit a region of low energy near 4, IC, = 120°, 130". These results839 might appear to be in conflict with the conformational maps for poly(Lproline),* since it is difficult to understand why a conformation should be available to poly(L-proline) but not to polysarcosine. Consequently the conformational maps for polysarcosine and poly(N-methyl-L-alanine)have been recomputed. The results previously reported for poly(N(1) Supported by Grant No. GB-36055 from the National Science Foundation. (2) W. L. Mattice, K . Nishikawa, and T . Ooi, Macromolecules, 6, 443. (1973). (3) W. L. Mattice and L. Mandelkern, J. Amer. Chem. Soc., 93, 1769 (1971). (4) D. A. Torchia, Macromolecules, 4,440 (1971). (5) D. A. Torchia, private communication. (6) The convention used is described in J. T . Edsall, P. J. Flory, J. C. Kendrew, A . M . Liquori, G. Nemethy, G . N. Ramachandran, and H. A . Scheraga, Biopolymers, 4, 130 (1966); J. Bioi. Chem., 241, 1004 (1966); J. Mol. Biol., 15, 339 (1966). In addition C,N will signify the carbon atom of the methyl group attached to the nitrogen atom of . refer ~ to rotation about the N,-CLN bond, which residue i and x ~ will is zero when the C t . ~ ' - N L and C,N-H bonds are eclipsed. The direction of rotation is identical for ~ 1 and , X~ N . ~ ( 7 ) N. GOand H. A. Scheraga, Macromolecules, 3, 188 (1970). ( 8 ) S. Tanaka and A. Nakajima, Polym. J., 1, 7 1 (1970). (9) J. E. Mark and M. Goodman, Biopolymers, 5,809 (1967).

methyl-~-alanine)gwere confirmed, but it was not possible to reproduce the conformational map reported for polysarcosine.8 In each case the conformational map is sensitive to the selection of the XN. If X N and are assigned so that the orientation of the methyl groups is similar to that of Cb and Cd in poly(L-proline), the computed energy us. $ for 4 near 120" does exhibit minima close to the locations found in the internal dipeptide of poly(L-proline).2

Calculational Procedure The atoms considered in the computation for poly(Nmethyl-L-alanine) are shown in Figure 1. The bond lengths and bond angles for poly(N-methyl-L-alanine)' were the same as those used by Mark and Goodman9 except when specifically stated to the contrary. In the computation for polysarcosine the three CB atoms and attached hydrogen atoms were replaced by three hydrogen atoms. The bond lengths and bond angles for polysarcosine were the same as those used by Tanaka and Nakajima.8 The nonbonded interaction was evaluated for each pair of atoms in Figure 1 whose distance of separation depends upon X L - I , XI,^, X I , ~ + I , X N . ~ ,X N , ~ + I , &, or +i utilizing the same functions used in the previous work.8>9The total number of interatomic distances used in the evaluation of the nonbonded energy was 171 for polysarcosine and 383 for poly(N-methyl-L-alanine).The torsional potentials about 4i and $ i were also the same as those used p r e v i o ~ s l y . ~The - ~ torsional barrier about X N was assumed to be identical with that about 4 ~ A. barrier height of 3 kcal/mol was used for torsion about ~ 1 . 1 0 Minima occur at f 6 0 and 180" for torsion about XN and XI. The electrostatic interaction of the peptide groups was computed by the same procedure used by Tanaka and Nakajima,8 which assigns partial charges to C', 0, N, and C N and assumes a dielectric constant of 3.5. The dihedral angles oi-l, ut,and I $ ~ +were ~ held constant a t 0"; qi-1 was also held constant a t 0" except when specifically stated to the contrary. (10) K . Nishikawa and T. Ooi, P r o p Theor. Phys., 46,670 (1971).